REAL backtesting isn't easy

Discussion in 'Strategy Development' started by Gordon Gekko, Sep 9, 2002.

  1. i've seen a lot of tradestation systems posted here and i use tradestation myself. i have come to the conclusion that most people aren't skilled enough to ACCURATELY test a system. i'm not just talking about slippage, etc. here. i'm talking about trade management. some people here may know what i mean. when i think of adding real trade management techniques to a tradestation system, MY MIND IS BLOWN. for this reason, i'm thinking of cancelling tradestation. not that tradestation isn't capable of doing it, i'd say most PEOPLE don't have the skills (including myself). however, it may still be useful for testing some strategies, but keep in mind it probably isn't a real world back test of how your trading system would/should be done. just my opinion....

    p.s. this may depend on the strategy being used.
  2. I was just thinking of trying Tradestation for their backtesting + data. Their large large data warehouse (accurate? or not?) seems to me to be a nice combination...

    However, would you do us all the favor of elaborating on your thoughts. What makes tradestation bad for trade management?

  3. ok, for example, what if you trade a system where you don't know at what price you will exit a bad trade? what if it is determined by 1 or more of a set of possible conditions? i'm not talking disaster stop here. every trade should have a disaster stop, yes. but, there are methods of trading in which you exit a trade because you were wrong, BUT YOU DIDN'T EXIT AT A PREDETERMINED STOP LOSS PRICE.

    have fun putting something like that accurately into tradestation. some people here may know what i'm talking about (darkhorse)...most won't.
  4. m_c_a98


    I still don't understand what type of stop orders you are talking about. Can you give another example?
  5. GG,

    Care to explain this a bit more?
  6. ok, this is a watered down version of what i'm talking about.

    say you're willing to hold overnight, but you make a rule that says you won't hold a trade overnight unless you have a 2% gain. if you don't have a 2% gain within 5 minutes of the close, you exit. you are exiting the position here, but not because it hit your stop loss point.

    now say you have more rules....or that the % gain required to hold overnight is dependent on the volatility of each stock. have fun programming all of this into tradestation. not saying it can't be done, but most people can't do it.

    look at tradestation strategies available on the internet.....they're mostly, "buy at this signal and sell at this signal." what about the other elements of managing the trades?
  7. m_c_a98


    All that stuff can be done. Look through the archives of the omega list.
    off the top of my head maybe something like:
    If time>=1555 and time<1600 then begin
    If marketposition=1 and close < entryprice(0)*1.02 then exitlong("l_exit");
    If marketposition=-1 and close > entryprice(0)-(entryprice(0)*.02) then exitshort("s_exit");
  8. Any data type strategy can be programmed, although I'm not saying it doesn't get complicated or that you have to be careful that you actually are getting what you think you are getting. When I used tradestation, I had a lot of trouble trying to program non-data type strategies, such as trendline breaks or divergences. I know there are ways to do it, but in practice it never seemed to work correctly all the time. Of course, Omega tech assistance was hopeless--sit on hold for 45 minutes to get some rep who had been hired two days ago and had no idea what you were talking about. Hopefully, all that is better now.
  9. I have heard that tradestation has no way of testing expectation, only win/loss ratio, since it doesn't include position sizing rules. Is that true ?
  10. you can export tradestation data to excel and run other types of analyses there...actually, you'll pretty much have to. tradestation will just give you the trades your system took. the information is just too basic.

    you might aswell just get the program that does it all in one - Wealth lab
    #10     Sep 13, 2002